A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document or patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever.
The invention relates to a complex, in crystalline form, of two proteins, FKBP12 and the FRB domain of FRAP, in association with rapamycin, a small organic molecule to which the proteins bind. The crystalline form of this ternary complex is particularly useful for the determination of the three-dimensional structure of the complex at the atomic level. The three dimensional structure provides information useful for the design of pharmaceutical compositions which inhibit the biological function of proteins such as FRAP which contain an FRB domain, particularly those biological functions mediated by molecular interactions involving rapamycin or other compounds capable of binding to an FRB domain.
Rapamycin (sometimes called sirolimus) was first described in 1975 as an antifungal agent isolated from Streptomyces hygroscopicus (Vezina, 1975; Sehgal, 1975). In 1987, the structurally related compound FK506 (sometimes called tacrolimus) was characterized as a potent immunosuppressive agent (Tanaka, 1987), and shortly thereafter, rapamycin was also shown to have potent immunosuppressive activity. In spite of rapamycin""s immunosuppressive activity and structural similarity to FK506, the two compounds suppress the immune response in completely different ways (Schreiber, 1992). FK506 inhibits the T cell receptor (TCR) signal and prevents activation of a resting helper T cell. Rapamycin inhibits the autocrine signaling pathway involving interleukin-2 (IL-2) and the IL-2 receptor (IL-2R). These latter signals commit the cell to a program of cell division by communicating with the components of the cell cycle machinery necessary for DNA replication.
Both FK506 and rapamycin are potentially useful in the treatment of human disease. FK506 has been approved by the FDA for use in treating the rejection of transplanted organs. A similar use has been envisioned for rapamycin, and its demonstrated activity in organ transplantation and autoimmune animal models indicate a high clinical potential. Rapamycin has been shown to have antitumor activity against B16 melanocarcinoma, colon 26 tumor, EM ependymoblastoma, CD8F1 mammary and colon 38 murine tumors (Sehgal, 1993). Rapamycin has also shown immunosuppressive activity in assays to measure prevention of development of autoimmune adjuvant arthritis, experimental allergic encephalomyelitis and autoimmune uveoretinitis in the rat (Sehgal, 1993).
The biological activity and structural novelty of both rapamycin and FK506 led to a search for their cellular target(s), and the target of both compounds was identified as the plentiful cytoplasmic protein FKBP12 (for FK506 binding protein) of 12 kDa molecular mass. Since FK506 and rapamycin bound to the same target (Kd of 0.4 and 0.2 nM, respectively) and affected different pathways, a new function was attributed to the FKBP12-ligand complex. This new function arises from the ability of FKBP12-FK506 and FKBP12-rapamycin complexes, but not the individual components, to bind to and inhibit still other protein targets. The FKBP12-FK506 complex inhibits the phosphatase activity of calcineurin, a crucial component of the TCR pathway. Calcineurin is a serine/threonine phosphatase also called PP2B. The FKBP12-rapamycin complex inhibits the IL-2R signal by binding to a large (289kDa) protein named FRAP in humans (Brown et al, 1994) or RAFT in rats (Sabatini et al, 1994; Chiu et al, 1994).
The structural basis for the tight binding of FK506 and rapamycin by FKBP12 has been investigated by both X-ray diffraction and NMR techniques (Clardy, 1995). In particular, high resolution X-ray structures are available for FKBP12-FK506 (1.4 xc3x85 resolution) and FKBP12-rapamycin (1.7 xc3x85 resolution) (Van Duyne et al, 1991; Van Duyne et al, 1991a; Van Duyne et al, 1993). These structures reveal, among other things, the fold of FKBP12, the atomic details of the hydrophobic binding pocket, and the details of how FK506 and rapamycin interact with the binding pocket. A structural analysis of the complex formed between FKBP12-FK506-calcineurin is also available (Griffith et al, 1995). That structure reveals how the portion of FK506 not involved in binding FKBP12 interacts with calcineurin and inhibits its phosphatase activity.
The biochemical characterization of FRAP, the target of the FKBP12-rapamycin complex, remains incomplete. The C-terminal domain resembles a phosphatidylinositol (PI) kinase, but to date no PI or protein kinase activity has been convincingly demonstrated. FRAP (RAFT, TOR) are members of a rapidly growing and important family of proteins that have been identified only recently (Zakian, 1995). ATM, TEL1, DNA-PK and MEC1 are some of the recently characterized members of this family of PIK-related kinases. (See e.g., Keith, 1995). ATM (for ataxia telngiectasia mutant) is responsible for a human autosomal hereditary disease characterized by cerebellar degeneration, progressive mental retardation, uneven gait, dilation of blood vessels, immune deficiencies, premature aging and a hundredfold increase in cancer susceptibility (Zakian, 1995). Persons who are heterozygous in ATM are believed to be at elevated risk for cancer. Mutations to TEL1 lead to abnormally short telomeres, and in conjunction with other mutations can lead to sensitivity to X-rays, UV radiation and hydroxyurea. DNA-PK is, as the name suggests, a DNA-dependent protein kinase that recognizes damaged DNA, and human cells without DNA-PK activity are radiation sensitive and repair deficient. MEC1 is required for both S-M and G2-M checkpoint progression as well as for meiotic recombination in yeast. Thus MEC1 is arguably the master checkpoint gene in yeast.
FRAP is a large protein (2549 amino acid residues), and only a small fraction can be involved in recognizing the FKBP12-rapamycin complex. Fortunately all of these residues are in one domain, and this domain, which is called the FKBP12-rapamycin binding (FRB) domain, is the protein used in this invention. It was identified through tryptic digests of FRAP and independently produced as an 11 kDa soluble protein (Chen et al, 1995)
Unfortunately, until now, three-dimensional structural details of the association of FKBP12-rapamycin with the FRB domain of FRAP have remained completely unknown. In the absence of such three-dimensional structural details, it has been impossible to design compounds based on that structure which would be capable of mimicking rapamycin""s binding to the FRB domain. We have now obtained crystals of that ternary complex and have determined its three dimensional structure. With this information, it is now possible for the first time to rationally design compounds capable of binding to an FRB domain and mimicking the pharmacological activity of rapamycin. Such mimics may be used in place of rapamycin as immunosuppressive agents or in other pharmacological applications.
This invention centers on the FRB domain of human FRAP and begins with obtaining crystals of human FKBP12-rapamycin-FRB of sufficient quality to determine the three dimensional (tertiary) structure of the complex by X-ray diffraction methods.
In considering our work, it should be appreciated that obtaining protein crystals in any case is a somewhat unpredictable art, especially in cases in which the practitioner lacks the guidance of prior successes in preparing and/or crystalizing any closely related proteins. Obtaining our first crystals of the ternary complex was therefore itself an unexpected result. In addition, our data represents the first detailed information available on the three dimensional structure of FRAP or of any of the PIK-related kinases and revealed an unpredicted array of surface features.
Our results are useful in a number of applications. As previously mentioned, the atomic details of how the FKBP12-rapamycin complex interacts with the FRB domain is essential for the structure-based design of rapamycin analogs. As noted above, rapamycin has several promising clinical indications, and improved rapamycin analogs would be useful therapeutic agents. This structure can be used as an essential starting point in predicting, via homology modeling, the structures of related proteins which contain homologous FRB domains, including other members of the PIK-related kinase family.
Furthermore, the structure showsxe2x80x94in atomic detailxe2x80x94how a small organic molecule, rapamycin, can be used to hold two proteins, FKBP12 and FRB, in close proximity. As such, this structure contains important lessons for the design of heterodimerizing agents.
Thus, the knowledge obtained concerning the FRB of FRAP can be used to model the tertiary structure of related proteins. By way of example, the structure of renin has been modeled using the tertiary structure of endothiapepsin as a starting point for the derivation. Model building of cercarial elastase and tophozoite cysteine protease were each built from known serine and cysteine proteases that have less than 35% sequence identity. The resultant models were used to design inhibitors in the low micromolar range. (Proc. Natl. Acad. Sci. 1993, 90, 3583). Furthermore, alternative methods of tertiary structure determination that do not rely on X-ray diffraction techniques and thus do not require crystallization of the protein, such as NMR techniques, are simplified if a model of the structure is available for refinement using the additional data gathered by the alternative technique. Thus, knowledge of the tertiary structure of the FRB region of FRAP provides a significant window to the structure of other proteins containing a homologous FRB domain, including the other PIK-related kinases.
Accordingly, one object of this invention is to provide a composition, in crystalline form, comprising a protein containing an FRB domain. The protein may have a bound ligand or may be part of a complex with a second protein molecule and a shared ligand. For instance, the crystalline composition may contain a complex containing a first protein having a peptide sequence derived or selected from that of an FKBP12 protein, e.g., human FKBP12; a second protein having a peptide sequence derived or selected from that of an FRB domain of a PIK-related kinase family member, e.g. the FRB domain of human FRAP; and a ligand such as rapamycin which is capable of binding to both proteins to form a ternary complex. Such a crystalline composition may contain one or more heavy atoms, e.g., one or more lead, mercury, gold and/or selenium atoms. Such a heavy atom derivative may be obtained, for example, by expressing a gene encoding the protein of interest under conditions permitting the incorporation of one or more heavy atom labels (e.g. as in the incorporation of selenomethionine), reacting the protein with a reagent capable of linking a heavy atom to the protein (e.g. trimethyl lead acetate) or soaking a substance containing a heavy atom into the crystals.
Preferred crystalline compositions of this invention are capable of diffracting x-rays to a resolution of better than about 3.5 xc3x85, and more preferably to a resolution of 2.7 xc3x85 or better, and are useful for determining the three-dimensional structure of the material. (The smaller the number of angstroms, the better the resolution.)
Crystalline compositions of this invention specifically include those in which the crystals are characterized by the structural coordinates of the FRB protein set forth in the accompanying FIG. 4 or characterized by coordinates having a root mean square deviation therefrom, with respect to backbone atoms of amino acids listed in FIG. 4, of 1.5 xc3x85 or less. Furthermore, our crystalline compositions include crystals characterized by the structural coordinates of both the FRB and FKBP12 proteins set forth in FIG. 4, optionally including a molecule of rapamycin as defined structurally by the accompanying coordinates therefor.
Structural coordinates of a crystalline composition of this invention may be stored in a machine-readable form on a machine-readable storage medium, e.g. a computer hard drive, diskette, DAT tape, etc., for display as a three-dimensional shape or for other uses involving computer-assisted manipulation of, or computation based on, the structural coordinates or the three-dimensional structures they define. For example, data defining the three dimensional structure of a composition of this invention or a portion thereof containing an FRB domain-containing protein of the PIK-related kinase family, or portions or structurally similar homologues of such proteins, may be stored in a machine-readable storage medium, and may be displayed as a graphical three-dimensional representation of the protein structure, typically using a computer capable of reading the data from said storage medium and programmed with instructions for creating the representation from such data. This invention thus encompasses a machine, such as a computer, having a memory which contains data representing the structural coordinates of a crystalline composition of this invention, e.g. the coordinates set forth in FIG. 4, together with additional optional data and instructions for manipulating such data. Such data may be used for a variety of purposes, such as the elucidation of other related structures and drug discovery.
A first set of such machine readable data may be combined with a second set of machine-readable data using a machine programmed with instructions for using the first data set and the second data set to determine at least a portion of the coordinates corresponding to the second set of machine-readable data. For instance, the first set of data may comprise a Fourier transform of at least a portion of the coordinates for the complex set forth in FIG. 4, while the second data set may comprise X-ray diffraction data of a molecule or molecular complex.
More specifically, one of the objects of this invention is to provide three-dimensional structural information on the FRB domain of FRAP, of other members of the PIK-related kinase family which containg homologous FRB domains, and of homologs or variants thereof, preferably in association with a bound ligand or bound ligand:protein complex (such as FKBP12-rapamycin). To that end, we provide for the use of the structural coordinates of a crystalline composition of this invention, or portions thereof, to solve, e.g. by molecular replacement, the three dimensional structure of a crystalline form of another such protein, protein:ligand complex, or protein:ligand:protein complex. Doing so involves obtaining x-ray diffraction data for crystals of the protein or complex for which one wishes to determine the three dimensional structure. Then, one determines the three-dimensional structure of that protein or complex by analyzing the x-ray diffraction data using molecular replacement techniques with reference to the previous structural coordinates. As described in U.S. Pat. No. 5,353,236, for instance, molecular replacement uses a molecule having a known structure as a starting point to model the structure of an unknown crystalline sample. This technique is based on the principle that two molecules which have similar structures, orientations and positions in the unit cell diffract similarly. Molecular replacement involves positioning the known structure in the unit cell in the same location and orientation as the unknown structure. Once positioned, the atoms of the known structure in the unit cell are used to calculate the structure factors that would result from a hypothetical diffraction experiment. This involves rotating the known structure in the six dimensions (three angular and three spatial dimensions) until alignment of the known structure with the experimental data is achieved. This approximate structure can be fine-tuned to yield a more accurate and often higher resolution structure using various refinement techniques. For instance, the resultant model for the structure defined by the experimental data may be subjected to rigid body refinement in which the model is subjected to limited additional rotation in the six dimensions yielding positioning shifts of under about 5%. The refined model may then be further refined using other known refinement methods.
For example, one may use molecular replacement to exploit a set of coordinates such as set forth in FIG. 4 to determine the structure of a crystalline co-complex of the FRB domain, FKBP12 and a ligand other than rapamycin. Likewise one may use that same approach to determine the three dimensional structure of a complex of FKBP12, rapamycin and a protein containing a modified FRAP FRB domain or an FRB domain from a homolog of FRAP.
Another object of the invention is to provide a method for determining the three-dimensional structure of a protein containing an FRB domain, or a complex of the protein with a ligand therefor, using homology modeling techniques and structural coordinates for a composition of this invention. Homology modeling involves constructing a model of an unknown structure using structural coordinates of one or more related proteins, protein domains and/or subdomains. Homology modeling may be conducted by fitting common or homologous portions of the protein or peptide whose three dimensional structure is to be solved to the three dimensional structure of homologous structural elements. Homology modeling can include rebuilding part or all of a three dimensional structure with replacement of amino acids (or other components) by those of the related structure to be solved. The structural coordinates obtained for the related protein or complex may be stored, displayed, manipulated and otherwise used in like fashion as those for the ternary complex of FKBP12-rapamycin-FRB set forth in FIG. 4.
Crystalline compositions of this invention thus provide a starting material, and their three dimensional structure coordinates a point of reference, for use in solving the three-dimensional structure of other proteins containing an FRB domain homologous to that of FRAP, as well as complexes containing such a protein. Sequence similarity may be determined using any conventional similarity matrix. (See e.g. Dayhoff,1979; Greer, 1981; and Gonnet, 1992). Proteins containing at least one FRB domain having at least 15% peptide sequence identity or similarity with respect to our FRB, as determined by any of the approaches described above, are considered FRAP homologs for the purpose of this disclosure.
By way of further example, the three dimensional structure defined by the machine readable data for the FRB domain (with or without the FKBP12 component) may be computationally evaluated for its ability to associate with various chemical entities. The term xe2x80x9cchemical entityxe2x80x9d, as used herein, refers to chemical compounds, complexes of at least two chemical compounds, and fragments of such compounds or complexes.
For instance, a first set of machine-readable data defining the 3-D structure of FRAP or a FRAP homolog, or a portion or complex thereof, is combined with a second set of machine-readable data defining the structure of a chemical entity or moiety of interest using a machine programmed with instructions for evaluating the ability of the chemical entity or moiety to associate with the FRAP or FRAP homolog protein or portion or complex thereof and/or the location and/or orientation of such association. Such methods provide insight into the location, orientation and energetics of association of protein surfaces with such chemical entities.
Chemical entities that are capable of mimicking rapamycin""s ability to associate with FRAP or a FRAP homolog should share part or all of rapamycin""s pharmacologic activities, e.g. immunosuppressive activity, but may be designed for more convenient or economical preparation, improved pharmacokinetics, reduced side effects, etc. Such chemical entities therefore include potential drug candidates.
The three dimensional structure defined by the data may be displayed in a graphical format permitting visual inspection of the structure, as well as visual inspection of the association of the protein component(s) with rapamycin or other chemical entities. Alternatively, more quantitative or computational methods may be used. For example, one method of this invention for evaluating the ability of a chemical entity to associate with any of the molecules or molecular complexes set forth herein comprises the steps of: (a) employing computational means to perform a fitting operation between the chemical entity and a binding pocket or other surface feature of the molecule or molecular complex; and (b) analyzing the results of said fitting operation to quantify the association between the chemical entity and the binding pocket.
This invention further provides for the use of the structural coordinates of a crystalline composition of this invention, or portions thereof, to identify reactive amino acids, such as cysteine residues, within the three-dimensional structure, preferably within or adjacent to a ligand binding site; to generate and visualize a molecular surface, such as a water-accessible surface or a surface comprising the space-filling van der Waals surface of all atoms; to calculate and visualize the size and shape of surface features of the protein or complex, e.g., ligand binding pockets; to locate potential H-bond donors and acceptors within the three-dimensional structure, preferably within or adjacent to a ligand binding site; to calculate regions of hydrophobicity and hydrophilicity within the three-dimensional structure, preferably within or adjacent to a ligand binding site; and to calculate and visualize regions on or adjacent to the protein surface of favorable interaction energies with respect to selected functional groups of interest (e.g. amino, hydroxyl, carboxyl, methylene, alkyl, alkenyl, aromatic carbon, aromatic rings, heteroaromatic rings, etc.). One may use the foregoing approaches for characterizing the FRB domain-containing protein and its interactions with moieties of potential ligands to design or select compounds capable of specific covalent attachment to reactive amino acids (e.g., cysteine) and to design or select compounds of complementary characteristics (e.g., size, shape, charge, hydrophobicity/hydrophilicity, ability to participate in hydrogen bonding, etc.) to surface features of the protein, a set of which may be preselected. Using the structural coordinates, one may also predict or calculate the orientation, binding constant or relative affinity of a given ligand to the protein in the complexed state, and use that information to design or select compounds of improved affinity.
In such cases, the structural coordinates of the FRAP or FRAP homolog protein, or portion or complex thereof, are entered in machine readable form into a machine programmed with instructions for carrying out the desired operation and containing any necessary additional data, e.g. data defining structural and/or functional characteristics of a potential ligand or moiety thereof, defining molecular characteristics of the various amino acids, etc.
One method of this invention provides for selecting from a database of chemical structures a compound capable of binding to FRAP or a FRAP homolog. The method starts with structural coordinates of a crystalline composition of the invention, e.g., coordinates defining the three dimensional structure of FRAP or a FRAP homolog or a portion thereof or a complex thereof. Points associated with that three dimensional structure are characterized with respect to the favorability of interactions with one or more functional groups. A database of chemical structures is then searched for candidate compounds containing one or more functional groups disposed for favorable interaction with the protein based on the prior characterization. Compounds having structures which best fit the points of favorable interaction with the three dimensional structure are thus identified.
It is often preferred, although not required, that such searching be conducted with the aid of a computer. In that case a first set of machine-readable data defining the 3D structure of a FRAP or FRAP homolog protein, or a portion or protein-ligand complex thereof, is combined with a second set of machine readable data defining one or more moieties or functional groups of interest, using a machine programmed with instructions for identifying preferred locations for favorable interaction between the functional group(s) and atoms of the protein. A third set of data, i.e. data defining the location(s) of favorable interaction between protein and functional group(s) is so generated. That third set of data is then combined with a fourth set of data defining the 3D structures of one or more chemical entities using a machine programmed with instructions for identifying chemical entities containing functional groups so disposed as to best fit the locations of their respective favorable interaction with the protein.
Compounds having the structures selected or designed by any of the foregoing means may be tested for their ability to bind to FRAP or a FRAP homolog, inhibit the binding of FRAP or a FRAP homolog to a natural or non-natural ligand therefor (e.g. FKBP12-rapamycin, in the case of FRAP), and/or inhibit a biological function mediated by FRAP or the FRAP homolog.
This invention also permits methods for designing a compound capable of binding to a FRAP or FRAP homolog based on the three dimensional structure of bound rapamycin. One such method involves graphically displaying a three-dimensional representation based on coordinates defining the three-dimensional structure of a FRAP or FRAP homolog protein or a portion thereof complexed with a ligand such as the FKBP12:rapamycin complex. Interactions between portions of ligand and protein are characterized in order to identify candidate moieties of the ligand for replacement. One or more portions of the ligand which interact with the protein may be replaced with substitute moieties selected from a knowledge base of one or more candidate substitute moieties, and/or moieties may be added to the ligand to permit additional interactions with the protein. Compounds first identified by any of the methods described herein are also encompassed by this invention.